Abstract
Expression patterns are a major feature of gene description, and are captured in many expression related databases (e.g., Bgee, Expression Atlas, GENEVESTIGATOR, or Tissue). Yet comparison and interoperability among these heterogeneous data sources or providers is limited by a lack of agreement for describing knowledge of gene expression. Ontologies are recognized as an interesting approach to achieve a consensus for knowledge representation, and allow to leverage semantic interoperability, but are lacking for the domain of gene expression. The Relational Ontology (RO) contains preliminary efforts to define terms for describing this domain. Notably, it contains `expressed in` and `expresses` properties (i.e., relations). Nonetheless, further work was necessary to address the following issue: the representation of additional information related to gene expression, such as developmental stages, as well as absent in and highly expressed relations. Furthermore, we reuse parts of the data schemas of the ORTH and UniProt core ontologies to provide the capacity to interoperate with other biological databases from different knowledge domains which are still relevant to the gene expression domain. Moreover, different gene expression databases use distinct confidence scores to assert expressed in or absent in relations. Finally, the GenEx ontology is fully adopted by Bgee, a gene expression database, as its core data schema to structure the Bgee RDF data accessible via its SPARQL endpoint at [https://www.bgee.org/sparql/](https://www.bgee.org/sparql/).